Cooperative Learning Intervention Decreases Influence of Deviant Peer Homophily on Likelihood of Friendship

• In Event: Using Social Network Analysis to Examine the Social and Organizational Networks of Children With Disabilities and Their Educators

Abstract

Perspective & Theoretical Framework
Adolescence is a critical window for adolescent biopsychosocial development. Decades of research has documented the influence of peers on adolescent deviant behavior, thus making adolescence a critical window for preventative interventions meant to mitigate negative peer influences. Van Ryzin & Roseth (2019) show a teacher-delivered, school-based cooperative learning intervention could increase peer relatedness in middle schoolers, leading to decreased trajectories of self-report deviant peer clustering, which then decreased trajectories of alcohol use over the course of two years.

Purpose
Despite gathering sociometric data—asking students to list six friends—four times over two years, Van Ryzin & Roseth (2019) did not assess how the intervention changed the network over time. Herein I conduct a secondary data analysis to examine the influence of this cooperative learning intervention on social networks over time, determining if the intervention decreases the probability of a tie based on equivalent levels of deviant peer clustering. In other words, I assess if the intervention moderates the influence of deviant peer association homophily on probability of friendship.

Methods & Data
The current work leverages the data from Van Ryzin & Roseth (2019): a randomized-controlled trial following seventh graders from fifteen Pacific Northwest middle schools for two years. Sixty ERGMs were fit—fifteen schools over four waves—to examine the influence of homophily in deviant peer clustering on probability of a tie, controlling for homophily along lines of alcohol use, school-reported sex, and school reported racial/ethnic minority status. Four meta-analyses—one per wave—were used to pool deviant peer association homophily effects on probability of reporting friendship for all schools.

Results
Significant heterogeneity was observed at all waves after baseline, according to significant Q statistics. Meta-regressions found the intervention did not explain a significant amount of heterogeneity until the final wave. In wave 4, intervention schools displayed a 33% lower chance of a tie when adolescents displayed homophily on deviant peer association, compared to control schools (OR = 0.67, p = 0.019). All waves demonstrated residual heterogeneity after inclusion of intervention in the meta-regression; in wave 4, though, it accounted for over 46% of the residual heterogeneity.

Significance
By pooling effects of the intervention on deviant peer clustering across schools, we show homophily in deviant peer association has less of an effect on friendship in intervention schools than in control schools after two years. This compliments findings in Van Ryzin and Roseth (2019) by showing the influence of the intervention on network ties, rather than self-report measures. The data suggest cooperative learning interventions foster a connected environment where homophily in deviant peer association is not as important in friendship ties, compared to learning environments that do not actively support mixing of student-selected groups during the learning process. Such findings have implications in the classroom and in the development of future preventative interventions.

Author

• Christopher Loan, University of Oregon